Problem with calculating the negative are enclosed between a curve and the x-axis

Hey Guys,

I have a cyclic voltammetry curve which forms a pseudo-rectangle with one of the long sides above the x-axis and the other is below the x-axis. I want to calculate the positive and negative areas (the area between the positive part of the curve and the x-axis and the are between the negative part of the curve and the x-axis). For that, I use the commands shown below. However, when I compare the areas I get from my code to that I get from the electrochemistry software (EC-Lab) I found that the positive areas are quite matching, however, this is not the case with the negative area. I am getting 0.893 from the software for the positive area and 0.8897 from my code which seems close but for the negative area, I am getting 0.883539 from the software and 0.7927 from my code (around 10% difference). Is there a problem in my code or is it the trapz function that doesn't work well with negative values or is it a data problem (attached in an excel sheet) ?

Thanks in advance.

gt0 = y>0;
gt1 = y<0;
Positive_area(1) = (trapz(x(gt0), y(gt0)));
Negative_area(1) = (trapz(x(gt1), y(gt1)));

Answer 1

[Torsten]

Try

idx = y<=0;
jdx = y>=0;
xn = x(idx);
yn = y(idx);
xp = x(jdx);
yp = y(jdx);
[xn,I] = sort(xn);
yn = yn(I)
[xp,J] = sort(xp);
yp = yp(J);
figure(1)
plot(xn,yn)
figure(2)
plot(xp,yp)
trapz(xp,yp)
trapz(xn,yn)

[Omar Hassan]

Thanks a lot. It really does help improving the values. I just have a question .... when looking at the edges of the plots, there are alternating lines instead of a smooth rising or decaying lines .... won't this affect the resulting values ?

[Torsten]

Yes, it will affect the value slightly. But can you localize these x-values and "repair" your data in this respect ? Will be a quite tedious job, I guess.

[Omar Hassan]

It will be hard to localize them especially that they differ with the CV :(

Thanks again.

[Omar Hassan]

This is what happens when I sort the value with more steeper curves

[Torsten]

Then your measurement data are - eh, not the best.